Operations | Monitoring | ITSM | DevOps | Cloud

In a multi-cloud environment, each cloud platform brings its unique tech stack to record events, manage services, set up configurations, manage user access and permissions, etc. While this allows you to leverage the best-of-breed services from different cloud vendors, the complexity of this setup makes it challenging to detect and respond to anomalies across clouds in real-time.

Monitoring production systems often feels like searching for a moving needle in a constantly shifting haystack. At Sentry, our goal was to empower customers to move beyond traditional threshold and percentage-based alerting. We aimed to help them detect subtle and complex anomalies in their systems in near real-time. This post will detail how our AI/ML team developed a time series anomaly detection system using Matrix Profile and Meta’s Prophet.

Site reliability engineers (SREs) often face challenges in keeping an organization’s sites running smoothly as the complexity of distributed systems steadily increases. With the rise of microservices, cloud-native architectures, and massive data volumes, manual monitoring and troubleshooting are no longer sustainable. SREs must navigate hurdles like alert fatigue, incident response delays, and the constant pressure to maintain system reliability.

Organizations invest heavily in robust systems, talented personnel, and sophisticated tools to ensure smooth operations. Yet, small anomalies often escape attention—minor glitches in applications, occasional lags in processes, or subtle irregularities in performance metrics. These may appear insignificant, but when left unaddressed, they can cascade into significant disruptions, leading to operational inefficiencies, financial losses, and reputational damage.

With this blog post, we continue our quarterly “What’s New” series to inform a broader audience about the latest features and improvements made to VictoriaMetrics Anomaly Detection (or simply vmanomaly). This post covers Q3'24 progress along with early Q4 to accommodate a slight shift in the publishing schedule — why not take advantage of it? Stay tuned for upcoming content on anomaly detection.

When you investigate an incident, context is everything. Let’s say you’re working on-call and get pinged in the middle of the night. You open the alert and it sends you to a dashboard where you recognize a latency pattern. But is the spike normal for that time of day? Is it even relevant? Next thing you know, you’re expanding the time window and checking other related metrics as you try to figure out what’s going on. It’s not to say you won’t find the answers.